Business

Know the Business — Meta Platforms, Inc.

Meta is two businesses bolted to one balance sheet: a global ad-auction monopoly-of-attention generating a 52% operating margin, and a hardware/AI moonshot losing about $19B a year. What mostly determines value is not whether Family of Apps grows 15% or 22% — it is whether the $125–145B 2026 capex bill converts into wider ad-auction spread within roughly two years, or whether it just becomes depreciation. The market currently pays a market-multiple (P/E 28, EV/EBITDA 16) on consolidated numbers that under-state core ad economics and over-state forward free cash flow.

1. How This Business Actually Works

Meta is an ad-auction marketplace that monetizes user attention at near-zero marginal cost. A user opens Facebook, Instagram, Threads, or Messenger; in milliseconds the platform ranks thousands of advertiser bids by expected value (bid × predicted click × predicted conversion) and serves the winner. Revenue mechanically equals daily active people × ads per person × price per ad. FY2025 is the textbook decomposition: DAP grew 7% to 3.58 billion, ad impressions grew 12%, average price per ad grew 9% — multiplying to the reported 22% ad-revenue growth.

The cost structure is unusual in three ways and they matter for valuation. First, more than 80% of operating expense is fixed — engineers, data centers, content moderation — so incremental revenue drops to operating income at roughly 70 cents on the dollar in a normal year. Second, the cost mix is rapidly shifting from labor to compute: depreciation and amortization rose from $11.2B (FY2023) to $18.6B (FY2025) and is set to roughly double again as the FY2025 capex base depreciates. The January 2025 useful-life extension of servers to 5.5 years is the one accounting choice that has held reported margin where it is. Third, stock-based compensation runs at 10% of revenue ($20.4B in FY2025) — a real economic cost that GAAP earnings under-state by treating it as non-cash.

Bargaining power sits squarely with Meta against advertisers and the long tail of small content creators, but Meta is itself subject to two parties it does not control: Apple/Google as mobile-OS gatekeepers (the 2021 ATT change cost a multi-billion-dollar slug of revenue per Meta's own disclosure) and EU regulators, who in 2025 forced consent-based ad targeting and a "less personalized ads" option that Meta itself calls "less relevant and effective." Both rails are why Meta has spent the past three years rebuilding the targeting layer with AI — the capex story is also a moat-defense story.

2. The Playing Field

The peer set sorts into three groups: scaled incumbents that own the ad auction (Meta, Alphabet), sub-scale ad platforms that do not (Snap, Pinterest, Reddit), and an attention competitor on a different business model (Netflix). Meta sits at the high-margin/high-scale corner that nobody else in the listed peer set occupies — the only true global peer with comparable economics is private (ByteDance/TikTok).

Three reads from this set. Meta has the highest operating margin in scaled digital advertising — 41% consolidated, 52% at the FoA segment — beating Alphabet despite being roughly half its size, because Meta does not carry Google's hardware, Cloud, and Other Bets drag in segment economics. Capex intensity is the new differentiator: Meta is spending 35% of revenue on capex versus Alphabet's 23% and the ad-platform challengers' under 4%. That gap will compress reported FCF margin for the next two years even if the auction stays strong. Pinterest and Reddit show that sub-scale ad businesses can run high FCF margins precisely because they do not spend the capex bill — but they also lack the targeting and global reach to win premium budgets, which is why their pricing power and absolute dollars are tiny.

3. Is This Business Cyclical?

Cyclical, but in a specific way: demand for ads, not supply of users. Engagement (DAP) has compounded through every macro shock since IPO, including COVID, the 2022 advertiser pullback, and the 2024–2025 election noise. What moves is the price per ad and, with a one-quarter lag, reported operating margin. The case study is 2022: rate hikes plus Apple's iOS 14.5 ATT change broke conversion attribution for direct-response advertisers, ad impressions still grew but average price per ad fell about 16%, and operating margin collapsed from 40% (FY2021) to 25% (FY2022) even though daily active people kept rising.

Two structural cycles overlay the cyclical advertising one. Privacy-driven signal loss (ATT in 2021, EU consent regime in 2025) periodically resets the targeting layer and forces a multi-year rebuild — Meta is mid-cycle on the second of those. The AI-capex cycle is the newest and currently dominates the equity. Meta spent $69.7B on capex in FY2025 (up 87% YoY) and guides $125–145B for 2026, funded with $29.9B of new debt in FY2025 alone. Headline operating margin held at 41% only because the January 2025 useful-life extension of servers stretched depreciation; the depreciation bill is set to compound for the next 3–4 years regardless of what ad revenue does.

4. The Metrics That Actually Matter

For Meta, headline EPS is the least informative profit number on the page — the FY2025 effective tax rate jumped from 12% to 30% on a one-time valuation allowance under the One Big Beautiful Bill Act (management says it would have been 13% absent the charge). The metrics below predict whether the business is improving or breaking.

Operating cash flow has grown every year and reached $115.8B in FY2025 — the core ad machine is fine. But capex has grown faster, more than doubling between FY2024 and FY2025. FCF will not exceed FY2024 again until either ad revenue accelerates sharply or capex growth flattens. Both are within Meta's control, but the market is now paying for the bet rather than the realized result.

5. What Is This Business Worth?

The right lens is earnings power on the core ad-auction engine, separated from the moonshot R&D drag and the AI-capex investment phase. Consolidated P/E 28 and EV/EBITDA 16 mislead in both directions: too low because the FoA segment alone earns more than the consolidated total operating income; too high because trailing free-cash-flow yield (2.8%) reflects a capex bill that has not yet shown up as depreciation. The mistake to avoid is treating either the headline multiple or a pristine FoA-only multiple as the answer. Reality Labs is a real, recurring cash drain — not a one-time loss — and the AI capex is a real obligation, much of it now backed by debt.

The practical valuation read: at ~$660 per share and ~$1.67T market cap (at the time of these segment figures), the market is paying about 18x FY2025 FoA-segment operating income (pre-tax, pre-RL) — roughly in line with where a regulated-utility-like ad-monopoly should trade if you believe (a) ARPP keeps compounding at 10%+ for three more years and (b) the FY2025–2026 capex cycle is a one-time step-up that flattens by 2027. If either belief breaks, the stock is not cheap.

6. What I'd Tell a Young Analyst

Watch the gap between ad-impression growth and price-per-ad growth every quarter. Both positive and roughly balanced (FY2025: +12% and +9%) means the auction is healthy. If impressions outrun price for two straight quarters, the platform is selling more lower-quality inventory at a discount — typically the first signal that monetization is fraying. If both go negative, that is the 2022 setup repeating.

Reality Labs is not a free option. It costs roughly $14B per year after tax — call it ~$8 per share of earnings — and management has guided that 2026 losses will be similar to 2025. Anyone modeling Meta on consolidated EPS without explicitly subtracting that drag from "core" earnings is double-counting cost. Anyone modeling it on FoA-only earnings without subtracting it as a recurring cash claim is double-counting value.

The 2025–2026 capex bill is the thesis. Meta is spending more on capex ($69.7B) than it returns to shareholders in buybacks ($26.2B) for the first time in its history as a return-of-capital story. The depreciation step-up from this capex base will compress reported margins for years. The investable question is not whether Meta is dominant in advertising — it is — but whether the AI bet generates auction-spread expansion (wider ARPP × higher impressions × better measurement) fast enough to offset that depreciation. Track Advantage+/Lattice/GEM commentary on conversion-rate uplift; that is where you will see the payback first.

Headline EPS is broken for FY2025. The 30% effective tax rate is a CAMT-driven valuation allowance, not a recurring rate. Use management's guided 13–16% rate for FY2026 to normalize, and rebuild your own EPS line from operating income.

The market underestimates two things, in my view. First, the structural moat of running the auction on 3.58B daily active people across four apps — there is no listed challenger within an order of magnitude and the only true peer is private (ByteDance). Second, the optionality embedded in AI glasses if Meta becomes the default consumer interface for personal AI (Ray-Ban Meta daily users tripled YoY). The market overestimates two things. First, the speed at which capex translates to revenue — historical precedent (Reality Labs, 2017–present) is that Meta's moonshots run longer and lose more than initial guidance. Second, the durability of the current 41% consolidated operating margin once depreciation from the FY2025–2026 capex base fully flows through.

If you can only watch one number, watch FCF per share. It captured Meta's 2022 stumble in real time, captured the 2023–2024 recovery in real time, and is the cleanest single signal of whether the AI investment cycle is paying back. It fell in FY2025 for capex reasons, not auction reasons — and whether it rises again in FY2026 is the next material test.